Postglacial migration supplements climate in determining plant species ranges in Europe

Abstract

The influence of dispersal limitation on species ranges remains controversial. Considering the dramatic impacts of the last glaciation in Europe, species might not have tracked climate changes through time and, as a consequence, their present-day ranges might be in disequilibrium with current climate. For 1016 European plant species, we assessed the relative importance of current climate and limited postglacial migration in determining species ranges using regression modelling and explanatory variables representing climate, and a novel species-specific hind-casting-based measure of accessibility to postglacial colonization. Climate was important for all species, while postglacial colonization also constrained the ranges of more than 50 per cent of the species. On average, climate explained five times more variation in species ranges than accessibility, but accessibility was the strongest determinant for one-sixth of the species. Accessibility was particularly important for species with limited long-distance dispersal ability, with southern glacial ranges, seed plants compared with ferns, and small-range species in southern Europe. In addition, accessibility explained one-third of the variation in species' disequilibrium with climate as measured by the realized/potential range size ratio computed with niche modelling. In conclusion, we show that although climate is the dominant broad-scale determinant of European plant species ranges, constrained dispersal plays an important supplementary role.

Figure 1.

Distribution, postglacial accessibility and variation in species occurrences uniquely explained by accessibility or climate for selected species: (a) Filipendula ulmaria; 64%, 0%; (b) Ranunculus psilostachys; 14%, 45%; (c) Koenigia islandica; 73%, 0%. Current distribution (black dots), hind-casting-based estimate of LGM distribution (empty circles), overlap between the two (half-filled circles) and accessibility to postglacial colonization from the estimated LGM range (green to orange: high to low). Maps are in the ETRS 1989 Lambert Azimuthal Equal Area projection.

Figure 2.

Geographical patterns in the importance of accessibility and climate for species occurrences across Europe. Average variation in species occurrences uniquely explained by (a) accessibility , or (b) climate . (c) Percentage of species with > . All values were calculated across species with a positive model-averaged accessibility coefficient (n = 655) in each ca 50 × 50 km AFE cell. Maps are in the ETRS 1989 Lambert Azimuthal Equal Area projection.

Figure 3.

Variation in species occurrences uniquely explained by accessibility for species with different (a) life forms (F, fern; A, annual herb; P, perennial herb; S, shrub; T, tree), n = 595; (b) long-distance dispersal (LDD) potential (measured as the number of LDD vectors), n = 346; and (c) climate-zone associations (ALN, northern-alpine; ALS, southern-alpine; ATL, Atlantic; BOR, boreal; CON, continental; MED, Mediterranean; PAN, Pannonian) n = 655 (shown for species with positive model-averaged accessibility coefficients, β_MA(A)). In squared parentheses, identical letters indicate no significant difference between groups (p < 0.05; Mann–Whitney U-test, significance levels adjusted using Bonferroni correction). In round parentheses, percentage of all species in a given group with a positive β_MA(A), and summed Akaike weights for accessibility of ≥95%; identical letters indicate no significant differences in percentages among groups (tested using χ²-tests, see the electronic supplementary material, appendix S4).

Figure 4.

Relationship between the importance of accessibility for species occurrences and (a) species current range size, (b) estimated range shift since LGM, and (c) range filling. Unique for accessibility represent the variation in species occurrences uniquely explained by accessibility after controlling for the effect of climate (shown for all species, n = 1016). Linear and Gaussian local (loess, fitted with span = 0.75 and a quadratic term) regressions were fitted either for all species or only for species with a positive model-averaged accessibility coefficient (n = 655). (a) Dashed line with circles, all (loess: R² 0.23); solid line, β_MA(A) > 0 (loess: R² 0.29); (b) dashed line with circles, all (linear: R² 0.20); solid line, β_MA(A) > 0 (linear: R² 0.07); (c) dashed line with circles, all (loess: R² 0.35); solid line, β_MA(A) > 0 (loess: R² 0.37).